Image processing method, image processing apparatus, and data recording medium

An image processing method in which an image signal forming an image space comprising plural pixels arranged in matrix in horizontal and vertical directions is divided into image signals corresponding to plural macroblocks of a rectangular shape into which the image space is divided, and image signals of subblocks of each macroblock are coded for each subblock, said method comprising the steps of:

• rearranging horizontal pixel rows of an image signal of a macroblock so that a first field image formed by image signals on odd-numbered horizontal pixel rows is positioned at an upper side of the macroblock and a second field image formed by image signals on even-numbered horizontal pixel rows is positioned at a lower side of the macroblock;
• transforming one of the image signal of the macroblock in which rearrangement has been performed and an image signal of a macroblock in which rearrangement has not been performed, into frequency components by frequency transformation, for each of four subblocks positioned upper left, upper right, lower left, and lower right in the macroblock;
• generating prediction values of frequency components of a target subblock to be coded by referring to at least one of subblocks which have been already coded and are positioned in the vicinity of and above the target subblock, in the vicinity of and at the left of the target subblock, and in the vicinity of and at the upper left of the target subblock, respectively, in the image space, wherein employed as prediction values of frequency components of the target subblock is one of a first weighted average of frequency component obtained among first group of plural coded subblocks which are positioned above the target subblock and aligned in the vertical direction, and a second weighted average of frequency component obtained among a second group of plural coded subblocks which are positioned at the left of the target subblock and aligned in the vertical direction;
• and
• coding difference values between the frequency components and the prediction values of the target subblock when an absolute value of difference of direct current components of frequency components between the decoded subblock positioned above and the decoded subblock positioned at the upper left is smaller than an absolute value of difference of direct current components of frequency components between the decoded subblock positioned at the left and the decoded subblock positioned at the upper left, the prediction values of the frequency components of the target subblock are generated by referring to frequency components of the decoded subblock positioned at the left of the target subblock, and
• when an absolute value is difference of direct current components of frequency components between the decoded subblock positioned at the left and the decoded subblock positioned at the upper left is smaller than an absolute value of difference of direct current components of frequency components between the decoded subblock positioned above and the decoded subblock positioned at the upper left, the prediction values of the frequency components of the target subblock are generated by referring to frequency components of the decoded subblock positioned above.

An image processing method for decoding a coded image signal for each subblock in which coding including rearrangement of horizontal pixel rows and frequency transformation has been performed to an image signal forming an image space comprising plural pixels arranged in matrix in the horizontal and vertical directions for each of subblocks of respective macroblocks of a rectangular shape into which the image space is divided, and method comprising the steps of:

• generating prediction values of frequency components of a target subblock to be decoded by referring to at least one of a decoded subblock positioned above and adjacently to the target subblock, a decoded subblock positioned at the left of and adjacently to the target subblock, a decoded subblock positioned at the upper left of and adjacently to the target subblock, wherein when the prediction values of the frequency components of the target subblock are generated, it is decided that one of the decoded subblock positioned above and adjacently to the target subblock and the decoded subblock positioned at the left of and adjacently to the target subblock is refereed to, based on direct current components of frequency components of the decoded subblocks which are positioned above and adjacently to the target subblock, at the left of and adjacently to the target subblock, and at the upper left of and adjacently to the target subblock, respectively;
• adding signals obtained by data analysis of the coded image signal of the target subblock to the prediction values to generate frequency components of the target subblock;
• transforming the generated frequency components of the subblock into an image signal by inverse frequency transformation;
• combining image signals of subblocks which belong to the same macroblock in the image space correspondingly to subblock positions in the macroblock, to generate an image signal of the macroblock;
• and
• with respect to an image signal of a macroblock in which horizontal pixel rows have been rearranged when it has been coded so that a first field image formed by image signals on odd-numbered horizontal pixel rows is positioned at an upper side of the macroblock, and a second field image formed by image signals on even-numbered horizontal pixel rows is positioned at a lower side of the macroblock, performing rearrangement for restoration to the horizontal pixel rows so as to form a frame image comprising the first and second fields, and with respect to an image signal of a macroblock in which the horizontal pixel rows have not been rearranged, regenerating an image signal forming an image space comprising plural macroblocks without performing rearrangement for restoration to the horizontal pixel rows.

An image processing method for decoding a coded image signal for each subblock in which coding including rearrangement of horizontal pixel rows and frequency transformation has been performed to an image signal forming an image space comprising plural pixels arranged in matrix in the horizontal and vertical directions for each of subblocks of respective macroblocks of a rectangular shape into which the image space is divided, said method comprising the steps of:

• generating prediction values of frequency components of a target subblock to be decoded by referring to at least one of subblocks which have been already decoded and are positioned in the vicinity of and above the target subblock, in the vicinity of and at the left of the target subblock, and in the vicinity of and at the upper left of the target subblock, respectively in the image space, wherein employed as prediction values of frequency components of the target subblocks is one of a first weighted average of frequency component obtained among a first group of plural decoded subblocks which are positioned above the target subblock and aligned in the vertical direction, and a second weighted average of frequency component obtained among a second group of plural decoded subblocks which are positioned at the left of the target subblock and aligned in the vertical direction;
• adding signals obtained by data analysis of the coded image signal of the target subblock to the prediction values to generate frequency components of the target subblock;
• transforming the generated frequency components of the subblock into an image signal by inverse frequency transformation;
• combining image signals of subblocks which belong to the same macroblock in the image space correspondingly to subblock positions in the macroblock, to generate an image signal of the macroblock;
• and
• with respect to an image signal of a macroblock in which horizontal pixel rows have been rearranged when it has been coded so that a first field image formed by image signals on odd-numbered horizontal pixel rows is positioned at an upper side of the macroblock, and a second field image formed by image signals on even-numbered horizontal pixel rows is positioned at a lower side of the macroblock, performing rearrangement for restoration to the horizontal pixel rows so as to form a frame image comprising the first and second fields, and with respect to an image signal of a macroblock in which the horizontal pixel rows have not been rearranged, regenerating an image signal forming an image space comprising plural macroblocks without performing rearrangement for restoration to the horizontal pixel rows.